面向池塘养殖的鱼类摄食行为监测算法OA

China is a major aquaculture country,accounting for more than 70％of the global aquaculture output,with pond aquaculture being one of its primary forms.It provides a stable source of high-quality protein for the national population.However,the low efficiency of traditional aquaculture models and the lack of low-cost intelligent monitoring equipment have restricted the application and popularization of precision feeding technology.To address the above problems,based on the correlation laws between the feeding desire of fish schools in pond aquaculture and their spatial distribution,activity state and other characteristics,this study used sonar to monitor fish school echoes,and analyzed the characteristics of vertical position changes,aggregation density and activity level of fish schools in water through echo signal processing,so as to identify the feeding behavior states of fish schools as strong feeding,weak feeding or non-feeding.The fish school distribution characteristics were simulated by an acoustic echo model,and the dynamic feeding behavior was simulated with an improved Boids algorithm.K-means clustering was applied to divide the water body into different regions.The spatial position,aggregation degree and activity level of fish schools were comprehensively analyzed,and the feasibility and effectiveness of the proposed method were verified in a simulation environment.Finally,the effectiveness of the proposed detection algorithm was validated through actual tests.The results show that the recognition accuracy of the proposed method for fish school aggregation and activity states reaches 88.1％and 68.3％respectively,and the accuracy is improved by more than 16.8％after fusing spatial information.This research provides a reliable basis for the realization of adaptive feeding in pond aquaculture.